an expert’s perspective - corporation for national and ...€¦ · an expert’s perspective...
TRANSCRIPT
An Expert’s PerspectiveA conversation with experts on evidence and the
role of government
An Expert’s Perspective
Moderated byWendy Spencer, CEO, Corporation for National and
Community Service
Panelists includeRon Haskins, Co-Director of the Brookings Center on
Children and Families and Budgeting for National Priorities Project, The Brookings Institution
Dr. Kathryn Newcomer, Director of the Trachtenberg School of Public Policy and Public Administration, George Washington University
Kathy Stack, Vice President of Evidence-Based Innovation, Laura and John Arnold Foundation
Ron Haskins
Co-Director of the Brookings Center on Children and Families and Budgeting for National Priorities
Project, The Brookings Institution
Show Me the Evidence: The Culture of Evidence is Growing
Growth of an Evidence-Based Culture• Obama’s evidence-based initiatives; 1,400 local projects
• Clearinghouses with lists of evidence-based programs
• OMB emphasizes evaluation by agencies
• “Nudge” unit in White House
• Growth of Pay for Success; more than 30 projects
• Results First (Pew and MacArthur; Clearinghouse)
• Results for America
• Coalition for Evidence-Based Policy
• Foundation support: Pew; MacArthur; Grant; Arnold; Annie E.
Casey; Casey Family Programs; Edna McConnell Clark; Gates;
others
Click to edit Master title style
Dr. Kathryn Newcomer
Director of the Trachtenberg School of Public Policy and Public Administration, George Washington University
Click to edit Master title style
“Evidence-based Policy,”
“Data-Driven Decision-making”–
the New Normal?
Click to edit Master title styleSetting the Stage…
• Program Evaluation: application of systematic analytical (social science) methods to address questions about program operations and results, including:
– Performance Measurement (Monitoring): routine measurement of program inputs, outputs, intermediate/ long-term outcomes attributed to a program
Both involve measurement + judgment!!
Click to edit Master title styleWhat are the Opportunities for Evidence to Inform Policymaking?
• Analyses of “performance” data collected by agencies (or delegated service delivery agents such as grantees)
• Implementation, Outcome and Impact evaluations performed by other agents for agencies
• Manipulations of services in experiments by agencies – “behavioral economics”
• Syntheses or systematic reviews of impact evaluations by external agents, e.g. websites like “What Works”
• Other?
8
Click to edit Master title styleWhat are Challenges for Evidence to Inform Policymaking?
• Expectations– What constitutes evidence?
– How transferable is evidence?
– Do we underestimate the role played by the “impactees?”
– Where is the capacity to support both the demand and supply of evidence?
9
Click to edit Master title styleWhy isn’t There Agreement About the Quality of Evidence?
• Differing professional standards and “rules” or criteria for evidence, e.g., lawyers, accountants, engineers, economists
• Disagreements about methodologies within professional groups, e.g., RCTs
• The constancy of change in problems and the characteristics of the targeted impactees
Click to edit Master title style
11
We Overstate the Certainty of the Evidence we Can Collect
Perceptions of the certainty of “evidence” have changed.
Click to edit Master title styleWe Overstate the Ease of Flow of Evidence
It plays a wide (enough) causal
role
Study conclusion: It plays a causal role there
Policy prediction: It will play a causal role
here
Source: Cartwright, N. (2013). Knowing what we are talking about: why evidence doesn't always travel.
Evidence & Policy: A Journal of Research, Debate and Practice, 9(1), 97-112.
Click to edit Master title styleWhy don’t RCT results alone make for well-supported effectiveness predictions?
It will play a positive causal role here
It can play a positive causal role here
It plays a positive causal role there
(and there)
R
C
T
R
C
T
Source: Cartwright, N. (2013). Knowing what
we are talking about: why evidence doesn't
always travel. Evidence & Policy: A Journal of
Research, Debate and Practice, 9(1), 97-112.
Click to edit Master title styleWhat is needed for a well-supported effectiveness (impact) prediction?
It will play a positive causal role here
The support factors for it are w, y, z
? ? ? ?
We have w, y, zhere
? ? ?
It can play a positive causal role here
It can play It plays a the same positive causal
role as role there (and there there)
S R RD
? T C CY
U T T
Source: Cartwright, N. (2013). Knowing what we are talking about: why evidence doesn't always travel.
Evidence & Policy: A Journal of Research, Debate and Practice, 9(1), 97-112.
Click to edit Master title styleWe Underestimate the Role of Volition Among
Impactees and their Own Heuristics
EFFECTSStructural and long term change
PREMISES
Based on information and evidence we decide that...
IMPLEMENTATION PLAN
IF we implement successfully certain activities...
ISSUEOpportunity
NeedProblem
MECHANISM
…THEN we trigger desired behaviors and
processes in a target group...
INPUTSMoneyPersonnelOther resources
ACTIONSOperationsProceduresProcesses
OUTPUTSInfrastructureServicesInformationDisincentives IncentivesChoice architecture
DETAILED PREMISES
FactsResearch result
InformationEarlier
experiencesOpinions
UNDERLYING THEORIESOn what premises do we base our
decisions?
THEORY OF IMPLEMENTATIONHow do we want to use inputs to produce
desired outputs?
THEORY OF CHANGEHow will positive change be produced?
Reaction of subjects
CHANGE
…AND THAT leads to positive, sustainable
change
Context
Heuristics and biases
Click to edit Master title styleWe Overstate The Current Evaluation Capacity in the U.S. Federal Government
Click to edit Master title styleEvaluation Capacity = Both Demand and Supply
• Who is asking for the evidence?
• Is there a clear understanding between providers and requestors for what evidence is needed?
• Are there sufficient resources within agencies to respond to demand?
• What about the lack of interaction and synergies among the different potential providers of evidence - such as GPRA/GPRAMA reporting staff, internal evaluation staff, external evaluation contractors, SBST, data.gov teams, etc.!
Click to edit Master title styleCurrently Fragmented and Non-Communicative
Performance Measurement
Impact Evaluation
Behavioral Economics
Click to edit Master title styleTo Strategic Use and Synergies!
Evaluation
Performance Measurement
Impact Evaluation
Implementation Evaluation
Behavioral Economics
Click to edit Master title styleAnswers? Apply Program Evaluation Tools
• Apply an evaluative mind set as well as apply appropriate skills
• Key elements of evaluation practice of special use in providing and assessing evidence include:
– Develop more realistic Theories of Change
– Frame appropriate questions
– Match designs to questions
20
Click to edit Master title styleProgram Logic Models Typically Ignore Context
Focus of Evaluation
Formative (Process/Implementation) Evaluation
OutcomeEvaluation
ImpactEvaluation
Desired Impact Long-term Outcomes
IntermediateOutcomes
Short-term Outcomes
OutputsActivitiesResources/Inputs
WHYHOW
Click to edit Master title styleA More Elaborate Theory of Change?
Desired Impact Long-term Outcomes
IntermediateOutcomes
Short-term Outcomes
OutputsActivitiesResources/Inputs
Contextual Factors outside of the control of program staff affect the extent to which programs and policies achieve desired outcomes!
Click to edit Master title styleIdentifying Key Contextual Factors is Especially Important in Understanding Our Theory of Change
• Assumptions about linkages, e.g., e.g., the teachers received the appropriate training to implement the new curriculum
• “Unobservables” in clients or beneficiaries- like motivation to continue in the mentoring program
• External barriers like lack of public transportation for clients to get to the service delivery site
• And some may reflect forces of nature or changes in political will, e.g., poor weather limits the amount of nature walks that are supposed to be included in the summer camping experience for inner city youth
Click to edit Master title styleFrame Useful Evaluation Questions
• The type of question – and type of desired evaluation –matter for determining appropriate methods:
– Descriptive Questions – describe input levels, outputs, contextual variables and/or measurable outcomes
– Normative Questions – assess levels of compliance for outputs or outcomes with criteria in law, regulation, or equity norms
– Impact Questions – measures of effectiveness or “impact”
– Explanatory Questions – explain implementation (fidelity) and processes
Click to edit Master title styleMatch Design to Research Questions
Principles of Evaluation Design
1. Frame the most appropriate questions to address
2. Clear and answerable evaluation questions drive the design decisions
3. Design decisions are made to provide appropriate data and comparisons needed to address the evaluation questions
4. Decisions about measurement and design are made to bolster the methodological integrity of the results
5. During design deliberations careful consideration should be given to strengthening inferences about findings
6. Goal to report that evaluation design and reporting decisions were characterized by strong methodological integrity
Click to edit Master title styleExamples
Objective Illustrative Questions Possible Design#1: Describe
program
(or regulatory)
activities
How extensive and costly are the program activities?
How do implementation efforts vary across sites,
beneficiaries, regions?
Has the program been implemented sufficiently to be
evaluated?
Performance Measurement
Exploratory Evaluations
Evaluability Assessments
Multiple Case Studies
#2: Probe
targeting &
implementation
How closely are the protocols implemented with fidelity
to the original design?
What key contextual factors are likely to affect intended
outcomes?
What feasibility or management challenges hinder
successful implementation of the program?
Multiple Case Studies
Implementation or Process
evaluations
Performance Audits
Compliance Audits
#3: Measure
impact of policy
or regulation
What are the average effects across different
implementations of the intervention?
Has implementation of the intervention produced results
consistent with its design (espoused purpose)?
Is the implementation strategy more (or less) effective in
relation to its costs?
Experimental Designs/RCTs
Non-experimental Designs: Difference-in-difference,
Propensity score matching, etc.
Cost-effectiveness & Benefit
Cost Analysis
Systematic Reviews & Meta-
Analyses
#4 : Explain how/
why produce
(un)intended
effects
How/why did the program have the intended effects?
To what extent has implementation of the program had
important unanticipated negative spillover effects?
How likely is it that the program will have similar effects
in other communities or in the future?
Impact Pathways and Process
tracing
System dynamics
Configurational analysis,
Click to edit Master title style
Kathy Stack
Vice President of Evidence-Based Innovation, Laura and John Arnold Foundation
An Expert’s Perspective – Q&A
Moderated byWendy Spencer, CEO, Corporation for National and
Community Service
Panelists includeRon Haskins, Co-Director of the Brookings Center on
Children and Families and Budgeting for National Priorities Project, The Brookings Institution
Dr. Kathryn Newcomer, Director of the Trachtenberg School of Public Policy and Public Administration, George Washington University
Kathy Stack, Vice President of Evidence-Based Innovation, Laura and John Arnold Foundation
Thank you for joining us for the 2015 CNCS Research Summit:
Focus on Evidence!
To learn more about research, evaluation and evidence at CNCS
visit us at
nationalservice.gov/evidence